TECHNICAL Reports
2009|07
Converting the TheSoz to SKOS
Benjamin Zapilko, York Sure
kölkölölk
GESIS-Technical Reports 2009|07
Converting the TheSoz to SKOS
Benjamin Zapilko, York Sure
GESIS – Leibniz-Institut für Sozialwissenschaften 2009
GESIS-Technical Reports
GESIS – Leibniz-Institut für Sozialwissenschaften
Informationelle Prozesse in den Sozialwissenschaften
Lennéstraße 30
53113 Bonn
Telefon: (0228) 22 81 - 175
Telefax: (0228) 22 81 - 4175
E-Mail: [email protected]
ISSN:
Herausgeber,
Druck und Vertrieb:
1868-9051 (Online)
GESIS - Leibniz-Institut für Sozialwissenschaften
Lennéstraße 30, 53113 Bonn
1
Introduction
The Thesaurus for the Social Sciences (TheSoz) [1] serves as a crucial instrument for indexing the
content of documents in the databases SOLIS (Social Science Literature Information System) and
SOFIS (Social Science Research Information System), both are owned and maintained by GESIS Leibniz Institute for the Social Sciences [2]. It contains overall about 11,600 keywords and covers
all topics and sub-disciplines of the social sciences. Additionally terms from associated and related disciplines are included in order to support an accurate and adequate indexing process of
interdisciplinary, practical-oriented and multi-cultural documents.
Paying attention to recent developments in the fields of e-Science and the web of Linked Open
Data (LOD) it seems to be quite obvious that the TheSoz has to be made available in the web in a
compatible and machine-readable format for providing and sharing its relevant information with
a greater community. With the development of SKOS (Simple Knowledge Organization System)
[3] there is a “standard way to represent knowledge organization systems using the Resource
Description Framework (RDF)”1. Due to the use of RDF [4] information can be passed and re-used
in a very interoperable way. First attempts for modeling the TheSoz with SKOS have been made
since the beginning of 2009 due to the fact that a lot of organizations and libraries were bringing
their thesauri and vocabularies to the web in SKOS format. These developments can also be ob2
served in Germany where recently the Thesaurus for Economics (STW) of the ZBW has been
published as linked data [5]. After SKOS has been announced as a standard in August 2009 by
the W3C a stable SKOS version of the TheSoz has been worked on. The following article describes the transformation process of the TheSoz in detail and takes an outlook on possible future
developments.
2
Converting the TheSoz to SKOS
The transformation process was split up into three steps and the method followed the established
method described in [6]. It began with a conceptual analysis of the thesaurus. Based on this all
elements of the thesaurus have been mapped to adequate properties of SKOS. Special use cases in
this mapping process are described in detail. The technical conversion process itself is automated
and done via XSLT methods.
2.1
Thesaurus Analysis
The Thesaurus for the Social Sciences contains about 11,600 keywords, of which more than 7,750
are descriptors (authorized keywords) and about 3,850 are non-descriptors. Relationships between
these keywords are expressed as broader, narrower or related terms as well as there are also “use
instead” and “use combination” relations and their counterparts (“used for” and “used for
combination”). Additionally a classification hierarchy is provided and each thesaurus term is
dedicated to one or more classification terms. The TheSoz contains a special type of nondescriptor called “AD” (for alternative descriptor) which differs from the international standard
norms for thesauri3 and holds more than one “use instead” and/or “use combination” relation
at the same time. There are about 200 of such “AD” terms in the TheSoz.
1
2
3
http://www.w3.org/2004/02/skos/intro
http://zbw.eu/stw/
ISO 2788 (Guidelines for the establishment and development of monolingual thesauri) and ISO 5964
(Guidelines for the establishment and development of multilingual thesauri)
GESIS-Technical Reports Nr. 7
ƒ
Example: The term “AD alternative energy” contains the relations “USE INSTEAD renewable energy” as well as “USE COMBINATION alternative AND energy”. Both of them are
(a combination of) descriptors with the same relevance. In most cases terms of the type “AD”
describe generic terms which have different concrete meanings in specified sub-contexts.
This is expressed through the use of more than one “use instead” and/or “use combination”
relations for only one term.
2.2
Mapping to SKOS
Most of the thesaurus items could easily be mapped to adequate SKOS properties and classes (see
table 1) due to the broad consistency of the TheSoz to the standard norms for thesauri. Problems
could be observed when mapping special data items and/or relations which are not conform to
thesauri standards to SKOS. This problem is an already well-known issue in research context [5],
but due to the fact that SKOS is based on RDF it is possible to define own relations without
greater effort.
Table 1:
Data Item
DD
ND
AD
NT
BT
RT
USE
UF
USK
Mapping of TheSoz data items to SKOS
Feature / Function
Property / Class
Descriptor / Preferred Term
Non-Descriptor / Non-Preferred Term
Alternative Descriptor
Narrower Term
Broader Term
Related Term
Use X instead of Y
Opposite of USE: X used for Y
Use Combination X AND Y
skos:prefLabel
skos:altLabel
skos:altLabel
skos:narrower
skos:broader
skos:related
skos:prefLabel
skos:altLabel
skos:prefLabel; skos:Collection;
skos:member
skos:altLabel
UFK
Opposite of USK:
Used for X with Y
scope
notationcode
Scope Notes for specific terms
Numerical code of one or more items of
the classification hierarchy to which the
term is dedicated to
-
Editorial Notes are used to include relations skos:editorialNote
which would get lost due to the non-conform
modeling of some relations of the AD
descriptor
skos:scopeNote
skos:notation
In case of the TheSoz the “use combination” relation has been modeled via grouping the affected terms as multiple “skos:member” in a “skos:Collection” which again is included in one
“skos:prefLabel“. But as mentioned above the TheSoz also contains a special type of nondescriptor called “AD” which holds more than one “use instead” and/or “use combination”
relation at the same time. Modeling such a term to SKOS would invoke more than one
“skos:prefLabel” in one single concept. Therefore these relations were modeled backwards via
their “used for” and/or “used for combination” relations in the associated descriptors as it is
suggested in various discussions in the web, but a small loss of information could not be avoided
with this solution. To avoid a complete loss of this relevant information these relations were
Converting the TheSoz to SKOS
3
included in additional “skos:editorialNotes” until there is a satisfying way to model them correctly with SKOS.
The classification hierarchy of the TheSoz (originally stored in a separate file) could be mapped
to SKOS without problems (see table 2). The numerical code of each classification term which
appears in “rdf:about” as well as in “skos:notation” is the same code which is included in
“skos:notation” of each concept containing descriptors (see table 1). By referencing these notation codes via URIs a connection between descriptors and their according classification terms is
established.
Table 2:
Mapping of TheSoz classification hierarchy to SKOS
Data Item
code
description
child
parent
2.3
Feature / Function
Property / Class
Numerical code of the classification item
Name of the classification item
Child nodes of the current classification item
Parent node of the current classification item
skos:notation
skos:prefLabel
skos:narrower
skos:broader
Conversion Program
Based on the mappings in the paragraph above the conversion program was developed. The
technical conversion process is done by XSL transformations. The original digital format of the
TheSoz which was already encoded in XML was converted to SKOS RDF. Additionally to the
mapping of each data item each concept got its own URI which provides a persistent and unique
identification. This is a very important aspect for re-use, availability and linkage in the web and
possible collaborations. The SKOS version of the thesaurus contains two types of URIs, one for
the descriptors and non-descriptors of the thesaurus and one for the terms of the classification
hierarchy. This allows easy distinguishing between descriptors and classification terms by only
knowing the concept URI. After the transformation process the resulting SKOS version of the
TheSoz was tested and validated by various established validation services for RDF4 and SKOS5.
3
Future Developments
Although the current SKOS version of the Thesaurus for the Social Sciences represents a stable
and valid data set which contains all of the original content of the thesaurus there is room for
improvements and extensions coming up in the near future. Beside mandatory technical enhancements and an upcoming update of the content there are currently a few more issues being
discussed and planned at GESIS:
ƒ
Thesaurus Browser
According to the Thesaurus Browser which is already available at the social science portal
sowiport.de [7], a similar web representation of the SKOS version of the TheSoz is discussed.
Exporting the data as RDF should be possible and provided as well as annotating the html
pages with RDFa [8] elements.
4
www.w3.org/RDF/Validator/
5
www.w3.org/2004/02/skos/validation
GESIS-Technical Reports Nr. 7
ƒ
SPARQL Endpoint
A SPARQL [9] endpoint on top of sowiport.de is planned, not only for querying concepts and
relations of the TheSoz, but also for retrieving all data provided in sowiport.de.
ƒ
Multi-Thesauri Setting
GESIS holds a large network of crosswalks between over 20 vocabularies which was resulting
6
from the project KoMoHe . The relevant information represented in these crosswalks could (at
a first step partially) be modelled with the mapping properties of SKOS. This could establish a
multi-thesauri setting in SKOS which uses already existing crosswalks between the involved
thesauri. The participated thesauri should also be available in SKOS.
These further plans indicate major steps towards the web of Linked Open Data (LOD) which seems
to be a very relevant research field for GESIS in the next years.
7
The current SKOS version of the TheSoz can be found and downloaded at the GESIS website .
References
1. Thesaurus for the Social Sciences: http://www.gesis.org/en/services/tools-standards/socialscience-thesaurus/
2. GESIS - Leibniz Institute for the Social Sciences: http://www.gesis.org/
3. SKOS - Simple Knowledge Organization System: http://www.w3.org/2004/02/skos/
4. RDF - Resource Description Framework: http://www.w3.org/RDF/
5. Neubert, J.: Bringing the “Thesaurus for Economics” on to the Web of Linked Data. Linked
th
Data on the Web (LDOW) 2009. Workshop at the WWW 2009 Conference, April 20 , Madrid,
Spain (2009)
6. Assem, M. van, Malaisé, V., Miles, A., Schreiber, G.: A Method to Convert Thesauri to SKOS.
In: The Semantic Web: Research and Applications. pp. 95-106 (2006)
7. SOWIPORT: http://www.sowiport.de/
8. RDFa: http://www.w3.org/TR/xhtml-rdfa-primer/
9. SPARQL - Query Language for RDF: http://www.w3.org/TR/rdf-sparql-query/
6
Competence Center Modeling and Treatment of Semantic Heterogeneity”; funded by the German Federal
Ministry for Education and Research; http://www.gesis.org/en/research/programs-andprojects/information-science/project-overview/komohe/
7
http://www.gesis.org/dienstleistungen/tools-standards/thesaurus-sozialwissenschaften/